Close this search box.

AI & Data Best Practices

Intelligent Process Automation: The 4 Levels of AI-Enablement

Table of Contents

This post was featured in our Cognilytica Newsletter, with additional details. Didn’t get the newsletter? Sign up here

Solving Increasingly Harder to Automate Tasks

During the industrial revolution, automation enabled machines to take over the hard, repetitive tasks that humans did, and replaced them with much faster, more robust, and less tiring steel, steam, gear and engine-driven improvements. This allowed us to go from making clothing by hand to mass-producing it by the boatload. The information revolution similarly accelerated commerce and business by replacing the huge rooms of typewriters, file cabinets, and lots and lots of secretaries with computers that could operate faster, with more information, and tirelessly. But the never-ending progress of productivity in business and industry didn’t stop there — the Internet revolution allowed companies to engage with millions more customers, around the clock, and provide levels of service and response they had never been able to before. So, are we at the end of the continuing progression of productivity, or can businesses revolutionize themselves once more with new technology?

Today’s knowledge workers are like the office workers of yesterday. They spend their time in email, on the phone, in various desktop and online apps and websites dealing with customers, suppliers, employees, partners, and internal stakeholders. Much of the time is spent dealing with various systems to shuffle information from one place to another, or enter/manipulate data from one system to another. If you’ve ever dealt with a bureaucratic organization, such as your Department of Motor Vehicles, you’re experiencing the joys of dealing with a knowledge-based service economy. But it doesn’t need to be this way.

Much of the reason why organizations seem to see limited productivity from their office and knowledge workers is because information resides in multiple, different systems, in different formats, and with various processes that determine how information can flow from one place to another. One might have thought that the move to Application Programming Interfaces (APIs) and other computer-based technology systems might have solved this problem. Yet while APIs have simplified the technical aspect of moving information from one place to another (sometimes), it has not solved the problem of dealing with differences in information. These various differences require a human to understand when information is needed, how it has to be manipulated, and how to utilize it for whatever particular task is needed by the organization.

Why Robotic Process Automation is Not Enough

Into this space of aggregating, managing, and manipulating data from a wide variety of sources is emerging a new class of automated “machine”: Robotic Process Automation (RPA) tools. These robots act on behalf of, or in place of, their human counterparts to interact with existing, legacy systems in the enterprise or anywhere online. They mimic the behavior of humans so that the human can focus on more important tasks for the company, rather than say, copying information from a website into a spreadsheet.

Yet, while RPA is making significant improvements into company’s operations by replacing rote human activity with automated tasks, Artificial Intelligence (AI) is poised to give this new engine of productivity a gigantic boost. RPA tools get stuck when judgement is needed on what, how, and when to use certain information in certain contexts. What if systems can learn from its human supervisors about how to utilize that information? Systems that leverage machine learning (ML) to dynamically adapt to new information and data will shift these systems from mere robots that automate processes to Intelligent Process Automation (IPA) tools that can significantly impact the face of the knowledge worker economy. Or as McKinsey Consulting puts it, “In essence, IPA takes the robot out of the human.”


Intelligent Process Automation: The Next Step

Even traditional RPA tools tend to get tripped up when things deviate substantially from what has been recorded. In particular, there are times when the context of the page needs to be understood, and different actions taken depending on understanding the circumstances. For example, if transcribing medical information from one system to another, the use of one laboratory system over another depends on the sort of diagnosis or treatment. Machine learning and other AI approaches can help deal with these situations by using natural language processing on text or spoken words, use different determinations on next steps based on learned interactions and thus provide a certain level of reasoning and insight on the different paths that the automated system can take.

In addition, there are many times when information is incomplete, requires additional enhancement, or combination with multiple sources to complete a particular task. For example, patient data might have incomplete history that is not required in one system but required in another. Another example is customer information that needs augmentation from other systems to provide greater value.  Intelligent systems can work to build and maintain a more complete profile of a customer, patient, employee, partner, consumer, or other individual and company and use this knowledge to help fill gaps in information received by different sources. In this way, intelligent process automation systems can help eliminate many of the exceptions that require human handling of RPA systems.

Cognilytica has spent time analyzing the IPA market, and we’ve determined that key capabilities fall into a few levels of “AI-Enablement” as we define below:


Level 0: Enhanced RPA (not AI) Level 1: Language & Context Aware Level 2: Intelligent Process Awareness Level 3: Autonomous Process Optimization
  • Screen recorder with visual flow designer
  • Complex rule sets
  • Complex user interaction capabilities with keyboard, mouse, swipe, and behavior modeling
  • Use of natural language processing tools for text (OCR), speech, and other interaction
  • Virtual assistants to help with process development
  • Fix and validate data as necessary for context
  • Can deal with unstructured data and inputs
  • Automatically identify process flows in new systems (“process discovery”)
  • Anticipate and mitigate process flow exceptions
  • Understand UI changes & make dynamic process changes
  • Find and fix missing or incorrect data
  • Automatic process documentation
  • Suggest and make modifications to processes to improve overall flow
  • Learn from itself to figure out better ways to handle process flow
  • Automatic orchestration of multiple bots to optimize processes
Source: Cognilytica – Intelligent Process Automation Report (

Learn More about the IPA Market in Cognilytica’s IPA Market Report

Curious to learn more about the IPA market, the key vendors to pay attention to, and our predictions for a market that we’re forecasting to be over $8.3B in combined software and services revenue by 2023? Then check out the IPA Market Report on the Cognilytica Site — free for Cognilytica Access Subscribers, and available for purchase for non-subscribers. You can also listen to our podcast on this topic: AI Today Podcast #22: Intelligent Process Automation.

Advance your Skills with CPMAI AI Best Practices

Login Or Register


Register to View Event


Get The Intelligent Process Automation: The 4 Levels of AI-Enablement


AI Best Practices

Get the Step By Step Checklist for AI Projects


Login to register for events. Don’t have an account? Just register for an event and an account will be created for you!